Robustly learning a Gaussian: Getting optimal error, efficiently

Abstract

We study the fundamental problem of learning the parameters of a high-dimensional Gaussian in the presence of noise | where an "-fraction of our samples were chosen by an adversary. We give robust estimators that achieve estimation error O(ϵ) in the total variation distance, which is optimal up to a universal constant that is independent of the dimension. In the case where just the mean is unknown, our robustness guarantee is optimal up to a factor of p 2 and the running time is polynomial in d and 1/ϵ. When both the mean and covariance are unknown, the running time is polynomial in d and quasipolynomial in 1/ϵ. Moreover all of our algorithms require only a polynomial number of samples. Our work shows that the same sorts of error guarantees that were established over fifty years ago in the one-dimensional setting can also be achieved by efficient algorithms in high-dimensional settings.